Hideyuki Shimonishi

A Design and Implementation of OpenFlow Controller Handling IP Multicast with Fast Tree Switching

We present a design of an OpenFlow controller handling IP multicast and a method to switch a multicast tree to the other with little packet loss. Reducing packet loss during reconstruction of multicast tree triggered by a failure has been an important problem to make multicast communications more reliable. One approach for this problem is to use redundant trees for multicast trees, but algorithms to compute redundant trees require centralized computation therefore cannot be used in IP network. We used OpenFlow, which has features of centralized control and programmability, to realize redundant multicast trees in the network. In this paper, we present a design of an OpenFlow controller supporting IP multicast and a method to set up multiple multicast trees of one multicast group in the network for fast tree switching. To avoid delivering duplicate packets to the receivers, we assign an ID to each tree, embed it into multicast packets, and select the tree to deliver multicast packets in the sender switch by rewriting the packet header. We showed little packet loss when switching the multicast tree triggered by the failure using implementation of the OpenFlow controller by the evaluation experiment.

Improved data distribution for multipath TCP communication

Multi-homed environments are increasingly common, especially for mobile users. To efficiently utilize multiple access lines for single file transfer, multipath TCP communication methods have been proposed. A multipath TCP enables simultaneous distributed data transfer between two end-points on multiple TCP connections. However, these methods cannot fully utilize the available bandwidth of multiple paths because they do not properly consider the end-to-end delay of packet transmission, so out-of-order data arrival at a receiver causes a bottleneck in data sort operations. This problem is more severe in environments where the quality of each path is different or unstable, such as in wireless environments. To solve this problem, we propose a multipath TCP communication method that includes a data distribution method to enable in-order delivery at a receiver. We call this arrival-time matching load-balancing (ATLB). ATLB continuously calculates the delay of each path, including the TCP queuing delay at a sender and the network delay, and then sends a data segment through the TCP connection with the lowest end-to-end delay. Simulation results show that ATLB improves end-to-end throughput, especially in heterogeneous environments where the quality of paths differs. For example, ATLB enabled twice the throughput with the conventional multipath TCP. We also report performance evaluation results from our ATLB test bed system in a wireless network environment. Our ATLB test bed system was able to fully utilize the aggregate available bandwidth of unstable multiple wireless links.

Virtualized network infrastructure using OpenFlow

In this paper, we present an introduction to our architecture for a virtualized network infrastructure and explain its enabling technologies including OpenFlow and Network OS for integrated control plane. To enable easy innovation within network research area, infrastructure virtualization to share a physical infrastructure among researchers and an idea of having programmability in control plane have been paid much attention. For example, OpenFlow defines atomic behaviors for packet handing within each switching element and a protocol interface to manipulate such behaviors from a control server, thus users can program their own network behaviors by injecting their own control programs into the server. In the same way, other protocols may be combined for optical and wireless nodes, or for network measurement. In addition to network control, other components in an infrastructure, such as virtual machines or data storage, would be controllable in the same way. Therefore, by integrating such control interfaces and creating integrated control plane, we propose a basic idea for virtualized infrastructure on which infrastructure creators have their own virtual infrastructure instance coupled with their own control programs on their virtualized control plane. Since on the control plane, a variety of control programs of many creators are installed to control a variety of infrastructure components, and they sometimes cooperates for mashing-up, or are sometimes isolated for virtualization, the role of a platform software of the control plane can be complex but very important. In this paper, we propose a network OS as a platform for such integrated control plane. It can be regarded as an OpenFlow controller but is extended to control entire infrastructure including OpenFlow switches and other IT and network resources. We develop a service model associated with its data structure and control module, and a model to construct a virtual infrastructure by defining the connections among the services. Then, we briefly explain our research projects developed on the network OS.

Dynamic fair bandwidth allocation for DiffServ classes

The assured forwarding per hop behavior standardized by the IETF Differentiated Services working group provides four class-based differentiated IP services. In this service, however, unexpected service degradation may occur and differentiation among classes may be disordered if the network is designed to minimize over-provisioning or is under-provisioned. We therefore developed a packet scheduling scheme that dynamically allocates bandwidth to each class queue to guarantee the differentiation among classes under any traffic conditions. The scheme estimates the sum of CIRs (committed information rates), i.e. rate of the packets having lowest drop preference, of active flows in each class and initially allocates the link bandwidth according to the sum of CIRs. It allocates the excess bandwidth by using a combination of CIR-proportional allocation and equal-share allocation. The equal share part enables that the flows in best effort class or the flows having zero CIRs can utilize minimum share of the bandwidth. Our scheme also introduces a scalable scheduling technique to improve fairness among flows in the same class. We evaluate the proposed scheme and show that it makes DiffServ operations fairer under any traffic conditions.

TCP-AFEC: An adaptive FEC code control for end-to-end bandwidth guarantee

Recently TCP is often used for video streaming applications. Since TCP congestion control is designed to be responsible for varying network conditions, its sending rate adaptation results in throughput fluctuations and leads to poor video quality. To address the problem, a FEC scheme is useful. Since lost packets can be recovered by redundant packets and are concealed from TCP, TCP can maintain higher throughput. To effectively utilize the FEC scheme, the redundancy level has to be appropriately determined so that redundant packets do not waste network bandwidth. Although many dynamic FEC mechanisms are proposed, these mechanisms are not appropriate for TCP video streaming because they do not try to satisfy the required rate. In this paper, we propose a new adaptive FEC scheme called TCP-AFEC that optimizes extra bandwidth usage for the redundancy level. Simulation evaluations show that the proposed mechanism performs better than previous FEC control approaches in various network conditions.

Deployable multipath communication scheme with sufficient performance data distribution method

Multi-homed environments are increasingly common, especially for mobile users. And several multipath communication techniques have been proposed, such as multipath TCP communication techniques. Multipath TCP techniques have potential to fully utilize multiple paths, but, these methods have difficulties in implementation and deployment, because users need to modify their applications or operating system or both. In this paper, we propose simple multipath communication technique, which we call Arrival-Time matching Load-Balancing (ATLB). ATLB is designed to be easily implementable on several environments. Besides, we also introduce overlay network approach which implicitly provides parallel data transfer scheme to users. The ATLB continuously calculates transmission delays of each path, including TCP queuing delay at a sender and network delay, and then sends a data segment through the TCP connection with the lowest delay. Simulation results show that ATLB realizes sufficient performance even in heterogeneous environments where the quality of paths differs. Measurement results over our wireless LAN test-bed system suggest that ATLB can fully utilize the aggregate available bandwidth over unstable multiple wireless links.

Performance analysis of fast reservation protocol with generalized bandwidth reservation method

The FRP (fast reservation protocol) utilizes a unique feature of ATM (asynchronous transfer mode) technology. The FRP is supposed mainly to be applied to LAN interconnection. We provide an exact analysis for a class of the FRP to obtain the burst (a protocol data unit of FRP) level performance of the FRP. One of the main features of the ATM is that the transmission rate can be adjusted according to the network congestion states. In our modeling, this feature is incorporated in such a way that the bandwidth a source requests is reduced if an attempt to reserve the bandwidth is rejected by the network. A rationale behind this is that rejection of the bandwidth request indicates network congestion. Therefore, the request with the smaller bandwidth after the reservation failure enables better sharing of network resources, which results in a performance improvement. To solve our model, which contains a very huge number of system states, we introduce a new numerical approach which is an extension of Sumita and Riederss (1991) replacement process approach. Through numerical examples, the appropriate transmission rate control method is examined, and considerations on backoff time and overhead of RM cells are also presented.

A network processor architecture for flexible QoS control in very high-speed line interfaces

We developed a network processor architecture that can be used for very high-speed line interfaces of carrier-class backbone routers and switches. Because advanced queuing and packet scheduling mechanisms are implemented as a software routine without the need for any special hardware components, the architecture provides a flexible QoS control mechanism and enables effective header handling.

TCP Congestion Control Enhancements for Streaming Media

Video streaming, including VOD (Video on Demand) services, over the Internet has been rapidly becoming popular. Although RTP/UDP has been considered as a standard transport protocol for video streaming, TCP is already widely used for VOD services because of its flexible accessibility to user clients beyond firewalls or NATs. However, poor video quality, for example, frequent pause of playback, due to TCP congestion control has been pointed out. In this paper, we propose an enhancement for TCP congestion control algorithm, which we call TCP-AV, to realize stable video streaming using TCP. TCP-AV incorporates two principal mechanisms, (i) dynamic TCP parameter tuning to stabilize TCP throughput around the target rate, and (ii) temporal target rate reduction to avoid severe congestion. Simulation results show that TCP-AV provides better rate control for maintaining target rate, and thus better video quality, even when the network is shared by many co- existing flows. In our scenario for a metropolitan VOD service, TCP-AV accommodates roughly 3 times larger number of VOD flows compared to TCP-Reno.

An improvement of weighted round robin cell scheduling in ATM networks

Weighted round robin (WRR) is a common cell scheduling method in ATM switches. In this paper, we study the delay performance of WRR cell scheduling and show that burstiness in input traffic is a major factor in performance degradation. We propose a new WRR scheme: WRR with save and borrow (WRR/SB), that helps improving delay characteristics in real-time traffic, and we provide an analysis of WRR and WRR/SB that describes their delay performance. We show that WRR/SB can provide an improved maximum delay bound over that of WRR alone.